Titlebook: Data Profiling; Ziawasch Abedjan,Lukasz Golab,Thorsten Papenbrock Book 2019 Springer Nature Switzerland AG 2019

PALSY

Intellectual

Profiling Non-Relational Data,, semi-structured data such as XML and RDF and non-structured data such as text. In this chapter, we describe two types of solutions: those which apply traditional data profiling algorithms to new types of data and those which develop new approaches to profiling non-relational data.

Connotation

左右连贯

Conclusions,s for discovering unique column combinations, functional dependencies among columns, and inclusion dependencies among tables. While the focus of this book is on exact profiling of relational data, we provided a brief discussion of approximate profiling using data sketches and profiling non-relational data, such as text and graphs.

microscopic

Classify

obligation

Discovering Metadata,the data or dependencies among columns, can help understand and manage new datasets. In particular, the advent of “Big Data,” with the promise of data science and data analytics, and with the realization that business insight may be extracted from data, has brought many datasets into organizations’

CYN

剥皮

Single-Column Analysis,ingle-column profiling tasks that we describe in more detail in the first part of this chapter. The second part discusses technical details and usage scenarios for certain single column profiling tasks. We refer the interested reader to Maydanchik [2007], a book addressing practitioners, for further

审问,审讯

Dependency Discovery,. tables, respectively [Toman and Weddell, 2008]. If the UCCs, FDs, and INDs are known, data scientists and IT professionals can use them to define valid key and foreign-key constraints (e.g., for schema normalization or schema discovery). Traditionally, constraints, such as keys, foreign keys, and